A recent paper in Nature Genetics describes the use of DNA microarrays (chips) to identify a gene expression signature that can distinguish non-metastatic primary tumours both from metastatic tumours and from primary tumours that are destined to metastasise [Ramaswamy, S. et al (2003) Nature Genetics 33, 49-54 (Abstract)]. The first step was to use DNA microarrays carrying gene sequences representative of thousands of different human genes to analyse the gene expression profiles of 64 primary and 12 metastatic tumours. Computer analysis of these profiles enabled a set of 128 genes to be picked out whose relative expression levels reliably distinguished most primary tumours from metastatic ones. A few primary tumours were apparently misclassified as metastatic, but further experiments showed that primary lung tumours that carried the metastatic gene expression signature were associated with significantly shorter survival times in the corresponding patients. The authors suggest that primary tumours carrying the metastatic gene expression signature are in fact destined to metastasise; if this is correct, then a large number of cells in these tumours must be running a genetic programme associated with metastasis, a finding that is at odds with the general view that metastasis is initiated by a few rogue cells within an otherwise genetically benign primary tumour. The authors went on to use computer analysis to whittle down their initial set of 128 genes to a set of 17 that gave equally good discrimination. The signature represented by the expression of these genes was used successfully to indicate the likelihood that various different types of primary tumour would metastasise. An exception was diffuse large B cell lymphoma, probably indicating a difference between solid and haematological tumours in the genetic mechanisms associated with metastasis.

Comment: This paper represents the latest contribution to a growing volume of research on the use of microarrays in oncology. The field of microarray technology is still struggling with problems of reproducibility and comparability between different laboratories and techniques, but greater awareness of these potential pitfalls, and the recent introduction of agreed standards for reporting microarray data, should go some way towards overcoming these difficulties. The hope is that information gleaned from the global analysis of gene expression enabled by microarrays will lead to better diagnosis and the identification of new drug targets.